I have Beyma TPL-150H and a B&C 8PE21 mid, and want to mount the midrange from the back of the baffle and adjust the baffle thickness such that both drivers are time aligned.
This means aligning the woofer's voice coil with the AMT membrane:
But what exactly should I align?
The AMT membrane is in the middle of the light-grey part of the motor assembly.
Where is the woofer's voice coil? In the exact middle of the magnet? Looking at drawings online it seems in some the voice coil is positioned closer to front vs the middle of the magnet.
Sorry for the basic question.
Thank you!
This means aligning the woofer's voice coil with the AMT membrane:
But what exactly should I align?
The AMT membrane is in the middle of the light-grey part of the motor assembly.
Where is the woofer's voice coil? In the exact middle of the magnet? Looking at drawings online it seems in some the voice coil is positioned closer to front vs the middle of the magnet.
Sorry for the basic question.
Thank you!
I thought time-alignment was a good enough goal in itself. Have come across a number of knowledgeable folks referring to time alignment as an import goal. Maybe not?
What is the physical baffle transition you refer? Transition from what to what? Baffle step would be below the mid's range, currently at 300Hz.
What is the physical baffle transition you refer? Transition from what to what? Baffle step would be below the mid's range, currently at 300Hz.
Such as flush mounting (just for example). Some make stepped baffles while attempting to time align, but that is creating a new problem.
Ignoring it would not be good, because the relative phase needs to be taken into account. There are other ways to manage this, for example in many cases you can handle it electrically.
Even if you were to attempt to find the acoustic depth you would do it by acoustic measurement. Phase changes are going to vary the effective delay which means the distance may change with frequency, and there would not be a single correct distance. Once you've measured you should see just how much electrical attention is needed.
The woofer itself will have a time delay depending on its high frequency rolloff characteristics, typically caused by Le forming a low pass filter with Re, and the mass of the woofer acting as another pole. Your driver has Le = 0.5 mH with Re = 5.6 ohms, so the first order filter from the electrical part has a -3dB point of about 1800 Hz. We know first order low pass filters have 45 degrees of phase lag at cutoff, so from there (with the speed of sound at 343 metres / second) we can deduce the woofer's delay will be the same as setting it back 2.4 cm. Add the mechanical rolloff and you may not need to align the woofer at all. Of course measurements will help immensely.
Also, consider flush mounting the drivers: your high frequency response will be smoother.
Also, consider flush mounting the drivers: your high frequency response will be smoother.
Compression drivers are tricky to determine acoustic centers on. Just aligning VC planes isn't enough to go on. Problem is sound travels at different speeds through various materials, so your acoustic center will vary based just on this alone. You have to make impulse measurements to determine actual acoustic centers to physically time align your drivers. Anything else is just guesswork.
The JBL 2441 compression drivers have a silver line on the housing that mark the alignment spot. The Lambda TD15 woofers have an alignment line as well. The center, front to back, of the Heils mark the centers. Will all those lined up, the speakers disappear.
That's interesting ... I didn't know that any drivers actually 'marked' acoustic centers.
Even after 'acoustic center alignment', using passive XO's can enter all-sorts of phase shifts.
So, final speaker system phase responses can be all-over the place.
And, once again, the test equipment so many people don't have access to can be very limiting.
Even after 'acoustic center alignment', using passive XO's can enter all-sorts of phase shifts.
So, final speaker system phase responses can be all-over the place.
And, once again, the test equipment so many people don't have access to can be very limiting.
Good. So flush mounting is in. Danny Richie mounts the mids and woofers from the back of the (open) baffle and adjusts baffle thickness to time align and this seems a good idea.
My system uses digital xo/correction that enable digital time alignment. I thought physical time alignment would be a better solution, but seems not necessarily. Plus front flush mounting is easier, so seems like a no-brainer. I'll start with that, measure and will learn what's the misalignment I should correct for physically in the next iteration of my panels.
Great!
Will read the Altec application note tonight. Thanks for including it!
My system uses digital xo/correction that enable digital time alignment. I thought physical time alignment would be a better solution, but seems not necessarily. Plus front flush mounting is easier, so seems like a no-brainer. I'll start with that, measure and will learn what's the misalignment I should correct for physically in the next iteration of my panels.
Great!
Will read the Altec application note tonight. Thanks for including it!
When designing my DIY hybrid electrostats I wanted to time align the woofer and stat panel but I wasn't sure whether the woofer's acoustic center would occur at the voice coil or forward of the voice within the cone.
Ultimately I decided that physically aligning the woofer's voice coil and stat panel would be both impractical and not aesthetically pleasing so I opted to compensate with time delay on the panel, using my DSP crossover.
I still believe that physical alignment is better than compensating with time delay, but every design has it's compromises and delay was better than nothing.
Although my DSP has time delay, it doesn't measure and compensate it automatically, so I proceeded as follows:
1) Setup the mic about 2ft away and between the woofer and stat panel.
2) Fed in a test tone matching the crossover frequency.
3) Added delay on the panel until the RTA showed max SPL (i.e. max constructive interference).
It's actually easier discern correct phasing by reversing polarities and adjusting delay for min SPL.
Result: 0.33 ms delay on the panel corrected phasing at the crossover frequency.
Ultimately I decided that physically aligning the woofer's voice coil and stat panel would be both impractical and not aesthetically pleasing so I opted to compensate with time delay on the panel, using my DSP crossover.
I still believe that physical alignment is better than compensating with time delay, but every design has it's compromises and delay was better than nothing.
Although my DSP has time delay, it doesn't measure and compensate it automatically, so I proceeded as follows:
1) Setup the mic about 2ft away and between the woofer and stat panel.
2) Fed in a test tone matching the crossover frequency.
3) Added delay on the panel until the RTA showed max SPL (i.e. max constructive interference).
It's actually easier discern correct phasing by reversing polarities and adjusting delay for min SPL.
Result: 0.33 ms delay on the panel corrected phasing at the crossover frequency.
Like above I play a test tone at the effective XO freqency but align by ear, for maximum combined loudness. If I could guess at the acoustic centers I would start with that offset then shift one driver back-and-forth (sometimes reverse polarity on one). For high frequency XO I've found the method to be sensitive and repeatable to 1mm, and imaging to work.
I would advise that you approach "time alignment" in a more holistic manner, one that includes the crossover. Aligning the acoustic center of the drivers is not all that useful in the end when the crossover filters will impart much larger time offsets due to group delay effects. Also, the relative position of the ACs changes with listening angle a bit.
The proper way to do this, in my opinion, is to include both the driver positions (in the front to back direction) AND the crossover in a sort of design feedback loop for which you are trying to optimize the phase response and minimize delay differences. This is not easy because the group delay profile (GD vs frequency response) of the crossover filters is not frequency independent while physically moving the drivers' mounting positions is. The consequence is that, for the finished loudspeaker, it is practically impossible to achieve linear phase (constant delay).
Finally, attempting physical alignment to even-out or equalize delay is ignoring the research into how sensitive human hearing is to delay differences across the audio band from a loudspeaker. There is plenty of research into this topic, and it turns out that your bog standard loudspeaker design that use NO delay alignment falls below the audible thresholds by a comfortable margin. The exception is when high (higher than 5th or 6th) order filters are used, but this is not typical of e.g. a loudspeaker using a passive crossover. While you might be horrified to see plots of temporal distortion of for example a square wave as reproduced by a loudspeaker, you must understand that your brain has a wonderful capacity to figure out what is going on and completely overcomes the temporal smearing. For an interesting example of this phenomenon, check out this video:
If you REALLY want to have a time aligned loudspeaker then the only way to achieve that to a high degree is by using FIR filtering that equalizes the group delay. You can just buy a fancy processor to do this. Just measure your finished loudspeaker's GD and then apply the inverse delay curve. Voila, perfect linear phase (at least at one position in space
).
Last edited:
Yes, but you may get false positives that way. As delay is the rate of phase per frequency you will get best results covering a range around the cross.
Amazon sell a tweeter phase checker, (you get a CD with clicks on it, and a hand held thing that has LEDs to tell you if the tweeter is in phase or not, for about £10) wouldn't it be possible to move the tweeter forward untill it's just out of phase, then move it backwards untill it's just out of phase, then move it to the centre of those to posisions to get it time alined?
I agree. The phase will vary with frequency which can change the alignment depending on phase response across the entire xover overlap region.